Betting on evolutionary architecture
summary
A 15-year-old C++ SOA science publisher making $3B a year gets replaced using a walking skeleton and microservice approach; heavy XSLT transforms on AWS S3 cause 35s page loads, so the team experiments: bump instance size (35s → 6s), move compute to the data center (→4s), optimize transforms (→3.5s), then add a Varnish cache with preload to reach 35 seconds to 0.2 seconds. Tools: AWS, S3, Ansible/Boto, Jenkins, siege/Vegeta, Graphite/alerting.
AI-written summary · about
chapters (13)
- 0:00Introduction
- 2:53What is software architecture
- 3:59What happens if these are easy to change
- 7:39How do we make decisions
- 18:59Betting
- 26:28Atomic bomb
- 39:33Decision
- 39:45Hypothesis driven architecture
- 42:10Monitoring observability
- 44:20Continuous delivery
- 45:40Performance testing
- 49:15Science
- 56:07Summary
related talks
42mBetting On Evolutionary Architecture
52mBetting on Evolutionary Architecture
18mSoftware Architectures Gone Wild
49mLarge-Scale Architecture: The Unreasonable Effectiveness of Simplicity
51mMicroservices and Microfrontends: Scaling, Learning, Growing
01h 03mEvolutionary Software Architectures
04mEvolution of service deployment
21mAvoiding That $1M Dollar AWS Bill - Mark Michael, Glenn Oppegard